Like David Brooks, I think Mitt Romney's Medicare premium support plan is the best piece of policy his campaign has produced. For one thing, it doesn't appear to be impossible, as Romney's tax promises do. Nor is it ludicrously unrealistic, as is the case with his budget promises, which imply a 40 percent cut to everything but Medicare, Social Security, and defense.

But that's all damning with faint praise, and Romney's premium support proposal deserves better than that. Premium support is a reasonable idea that often gets presented in an unreasonable way -- typically in a way that assumes huge savings and puts seniors on the hook if they're not realized. In 2011, for instance, Paul Ryan's budget used premium support to hold Medicare spending to the rate of inflation, which is far, far lower than the rate of health-care spending. There's no way that's going to happen. The result would've been massive cost-shifting to seniors. 

But there's a reason Ryan and others have tried to tie premium support to unrealistic caps. There's not much evidence that it saves money.

You wouldn't know that from Brooks's column. At the top, he says of the president's plans to move Medicare toward pay-for-quality that "the current budget projections are so bad because almost no one outside the employ of the president believes this approach will reduce Medicare costs."

That's flatly untrue -- a large number of health-care experts are optimistic on our ability to use comparative effectiveness research, electronic medical records, and expedited reform procedures to bring down costs. What is true is that the folks who do the budget projections -- namely, the Congressional Budget Office -- say there's not enough evidence to be confident about the plan. But they'll tell you the exact same thing about premium support. That's why Ryan and others include their caps, and why Romney's plan, which doesn't actually have a cap, won't substantially change the budget projections if presented to the CBO. 

I don't know if premium support will work. For now, sign me up as skeptical. But I'm skeptical about any approach to health-care costs that doesn't include a lot of trial-and-error. The appropriate emotion when confronting health-care costs is humility. And I'm certainly willing to believe premium support might work, or might help.

But what worries me is something Brooks writes toward the end of his column: "At worst, if the market approach flopped, we'd be back to where we started." That's not right.

Medicare's great advantage is it's a huge and powerful purchaser. That's allowed it to negotiate very low rates on health services -- an equivalent Medicare plan in the private sector costs about 20 percent more -- and it gives it the power to make big changes in the health-care marketplace by throwing its weight around. For instance, pretty much every doctor in the country is moving toward electronic medical records right now. That's a long overdue change, but it's only happening because Medicare is forcing them to do it. 

Premium support, by its very nature, fractures Medicare among dozens of private plans. So Medicare loses its pricing and market-leading power. Those are big losses in the fight for cost control. If premium support fails, or simply doesn't totally work, and we want to use Medicare's pricing power or its ability to move providers, we need to somehow rebuild that capacity, and now we have to do it over the objections of rent-seeking private insurers and medical providers. 

That's why my preference would be to try premium support and the pay-for-quality schemes simultaneously. Premium support can be tested in the Affordable Care Act's exchanges, which use a competitive-bidding process identical to the one envisioned by Romney, and by bringing competitive bidding into the Medicare Advantage program, as the Urban Institute suggests. Meanwhile, traditional Medicare can keep forcing changes to the way providers deliver care -- a project they've already begun in earnest, and that shouldn't be interrupted.

If, in a few years, we see competitive bidding succeeding beyond our wildest expectations, and the pay-for-quality effort is failing, we can then rebuild Medicare around premium support. if both are succeeding, we can figure out a way to merge the two. But what if premium support fails?

My concern with the Romney proposal is that he repeals the Affordable Care Act, which means repealing all those Medicare reforms, and he breaks Medicare down into a new and untested program. Rather than trying everything and being ready to see some efforts flop, he's betting the entire future of the health-care system on the premium support. If that fails, as it very well might, he's actually made trying anything else much harder.